Strain controllable magnetocrystalline anisotropy in FeRh/MgO bilayers

Eun Sung Jekal; Henry Hoffmann

REVIEW 4 major objections 6 minor 4 references

Strain controllable magnetocrystalline anisotropy in FeRh/MgO bilayers

T0 review · 4 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A thin FeRh film on MgO can have its magnetic easy axis flipped by strain, because strain shifts the iron d-electron states that carry spin-orbit coupling.

desk verdict Plausible but unverifiable DFT prediction of strain-driven MCA switching in FeRh/MgO, undermined by missing data and a strain model that co-strains the substrate. read the letter →

arxiv 1908.04761 v1 pith:QXEP7XB4 submitted 2019-08-06 cond-mat.mtrl-sci

Henry Hoffmann , Eun Sung Jekal This is my paper · ORCID

classification cond-mat.mtrl-sci PACS 71.15.Mb75.30.Gw75.70.-i

keywords FeRhmagnetocrystallineanisotropyepitaxialstrainspin-orbitcouplingantiferromagneticspintronicsdensityfunctionaltheoryMgOsubstrateeasyaxisreorientation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper uses first-principles electronic structure calculations to show that epitaxial strain can reorient the magnetization direction of an antiferromagnetic FeRh thin film grown on MgO. In the antiferromagnetic phase, compressing the film by half a percent makes the magnetic easy axis lie in the film plane, while stretching it by half a percent makes the easy axis stand out of plane, with the switch occurring near zero strain. The authors trace this to spin-orbit coupling between specific iron d-orbitals at the Fe/MgO interface, whose energies shift relative to the Fermi level as strain changes. If true, this gives a practical, substrate-driven knob for tuning magnetic anisotropy in antiferromagnetic spintronic devices.

What carries the argument

The load-bearing object is the second-order perturbation theory expression for the magnetocrystalline anisotropy, evaluated from the spin-orbit coupling between occupied and unoccupied states of the interfacial Fe atoms. In the slab geometry, the relevant channel is the coupling matrix element $\langle d_{xz,yz}|L_{xy}|d_{x^2-y^2}\rangle$ between iron $d$ orbitals; strain shifts these states relative to the Fermi level at particular $k$-points, converting a net negative anisotropy into a positive one. The slab supercell itself, with five FeRh monolayers on a MgO substrate and only atomic $z$-positions relaxed, is what makes the epitaxial strain well-defined.

What would settle it

A direct DFT calculation of the same five-monolayer FeRh/MgO slab with the opposite (Rh-terminated) interface, or with full in-plane relaxation, that fails to show a magnetocrystalline anisotropy sign change between -0.5% and +0.5% strain would falsify the claim; so would an experimental torque or Kerr measurement on epitaxial FeRh/MgO under biaxial strain showing no easy-axis rotation in that strain window.

Watch

Extended reading notes

Core claim

The central claim is that the magnetocrystalline anisotropy of an ultrathin FeRh/MgO bilayer changes sign around zero epitaxial strain, so that the magnetic easy axis rotates from in-plane under compression to out-of-plane under tension. The sign change is driven by strain-induced shifts of the spin-orbit-coupled $d$ states of the interfacial Fe atoms: at compressive strain the $\langle d_{xz,yz}|L_{xy}|d_{x^2-y^2}\rangle$ coupling dominates with a negative contribution to the anisotropy energy, while under tensile strain band shifts suppress that negative channel and enable a positive one, reversing the easy axis. The authors further find that the G-type antiferromagnetic phase remains more stable than the ferromagnetic phase by about 20 meV per Fe atom across the strain range considered, so the reorientation is an antiferromagnetic effect. The claim is specifically for the slab model used: five FeRh monolayers on MgO(001) with an Fe-terminated interface and oxygen atoms sitting atop iron.

Load-bearing premise

The whole prediction rests on the slab model where FeRh is locked to the MgO lattice with an Fe-terminated interface and only atomic z-positions relax; if the real interface termination, film thickness, or strain relaxation differs, the strain-driven easy-axis switch could disappear or reverse.

Editorial extensions

If this is right

In the G-AFM phase, the easy axis can be selected between in-plane and out-of-plane by choosing compressive or tensile epitaxial strain, without changing the magnetic order.
The interface Fe atoms, not the surface Fe atoms, dominate the anisotropy, so interface chemistry is the control parameter.
The G-AFM phase stays about 20 meV per Fe atom more stable than the FM phase over the +/-0.5% strain window, putting the switch in a robust antiferromagnetic state.
Strain engineering, including growth on lattice-mismatched or piezoelectric substrates, becomes a viable route to set magnetization direction in FeRh-based spintronic devices.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The predicted sign of the strain effect likely depends on the Fe/O interfacial registry; a Rh-terminated interface or a different oxygen position could suppress or reverse the reorientation, which is a testable extension of the same calculation.
Because the reorientation is an interfacial effect, increasing film thickness should dilute the anisotropy change, so the practical strain window should shrink for films much thicker than five monolayers.
On piezoelectric substrates, dynamic strain could offer an in-operando switch of the easy axis, going beyond the static strains studied here.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a competent but incomplete DFT paper. The new piece is the systematic strain-dependent MCA calculation for an FeRh/MgO bilayer, with a band-resolved explanation in terms of SOC between Fe d-states. That is a reasonable extension of prior work on structural stability under strain (refs 16,17). The methodology is standard (PBE, PAW, thin slab) and the perturbation-theory analysis is post-hoc interpretation, not fitting. Significance is incremental — strain-controlled MCA is established in other systems — but for FeRh/MgO this specific new application is useful.

But the central claim cannot be checked from the text. Table I, Fig. 2, Fig. 3, and Eq. (1) are all referenced but missing. The abstract says the easy axis switches from perpendicular to in-plane, while the results section says compressive strain gives in-plane and tensile gives out-of-plane. One of those is backwards. That kind of inconsistency matters in a paper whose only quantitative output is the sign of MCA.

The bigger problem is the strain definition. The supercell has one in-plane lattice constant for FeRh and MgO. At s=0 the FeRh is at its bulk value and the MgO is stretched by +0.6%; at s=+0.5% the MgO is stretched by over 1%. A thick MgO substrate cannot be strained that way, so the calculated 'epitaxial strain' path is not physically realizable for FeRh/MgO. Since the O atoms at the interface belong to the strained MgO, the Fe-O distance changes with s, and the reported MCA switching could be driven by substrate strain rather than intrinsic FeRh d-band shifts. The natural FeRh/MgO mismatch sits at about -0.6% strain on FeRh, at the edge of the range. So the central prediction is unsupported unless the two components' strains are decoupled.

The paper does have useful pieces. The observation that the G-AFM phase is stable across the strain range and the strain dependence of the orbital moment differences is a sensible direction. The mechanism section, once the figures are restored, could be informative.

My recommendation: this is worth a serious referee, but not in its current form. The authors need to supply the missing table and equation, fix the direction language, and redo or reframe the strain calculations so that the film strain is varied while the substrate remains at its own equilibrium lattice constant. If that changes the result, so be it; the paper would at least be consistent.

Referee Report

4 major / 6 minor

Summary. The manuscript reports first-principles DFT (VASP, PBE, PAW) calculations of the magnetocrystalline anisotropy (MCA) of a five-monolayer FeRh(001) film on an MgO(001) substrate, for biaxial strain s in the range -0.5% to +0.5%. The authors claim that the easy axis switches from in-plane under compressive strain to out-of-plane under tensile strain, and that the effect originates from strain-induced shifts of the interfacial Fe d-states (dx2-y2 and dxz,yz), analyzed via second-order perturbation theory and k-resolved band character. They also report that the G-type antiferromagnetic phase remains more stable than the ferromagnetic phase by about 20 meV/Fe over the strain range.

Significance. If the central claim holds, the paper would provide a concrete strain-engineering route for controlling the magnetic anisotropy of antiferromagnetic FeRh films, which is of interest for AFM spintronics. The study uses standard DFT methods, contains no fitted parameters, and offers a specific orbital-level mechanism. However, the manuscript as provided is incomplete: the numerical MCA data (Table I), the perturbation-theory expression (Eq. 1), and the supporting band-structure/PDOS figures (Figs. 2-3) are referenced but not present, and the slab model applies strain to the entire FeRh/MgO stack, which may not correspond to a physically achievable epitaxial condition on a thick MgO substrate. These issues prevent verification of the central claims in the current version.

major comments (4)

[Section 3, Table I and Fig. 2] The central quantitative claim — that MCA changes sign between s=-0.5% and s=+0.5% — is not verifiable from the provided text, because Table I (with the MCA values and moments) and Fig. 2 (the k-resolved orbital character) are referenced but their content is absent. The paper should include the actual MCA values (with units of erg/cm2) for each strain and each magnetic phase, and the figure should be present.
[Section 2] The slab model uses a single in-plane lattice constant for the entire FeRh/MgO supercell, and the strain s is applied to the whole stack. At s=0, the FeRh in-plane constant is 2.995√2 = 4.235 Å, while bulk MgO is ~4.21 Å, so the MgO substrate is already at +0.6% tensile strain; at s=+0.5%, the MgO strain is ~+1.1%. A thick MgO substrate would not be strained in this way in an epitaxial FeRh/MgO bilayer. The natural FeRh/MgO mismatch is about -0.6% (FeRh in-plane compressed), which lies at the edge of the studied range. The authors should disentangle film strain from substrate strain, e.g., by fixing the MgO at its bulk lattice constant and varying the FeRh in-plane constant, or by explicitly justifying the co-strained model. Without this, the predicted spin reorientation near s=0 is not clearly established for real FeRh/MgO heterostructures.
[Abstract, Introduction, and Section 3] There is a direct contradiction about the direction of the easy-axis switching: the abstract states the switching is 'from perpendicular to in-plane', while the introduction and Section 3 state that compressive strain gives an in-plane easy axis and tensile strain gives an out-of-plane easy axis. Next to this, the text 'gives the negative MCA of 0.47 erg/cm2 for s=0.50%' appears inconsistent with the claimed sign of the effect; this likely should read 's=-0.50%' and '-0.47 erg/cm2'. The sign convention and the direction of switching must be made consistent throughout.
[Section 3, Eq. (1)] The perturbation-theory expression for the MCA is missing from the manuscript; the text refers to 'Eq. (1)' but the equation itself does not appear. Since the orbital-resolved analysis is entirely based on this expression, the equation must be provided.

minor comments (6)

[Throughout] Typographical errors: 'usingab-initio' (Abstract), 'antiferromagetic' (Abstract), 'expitaxially' (Introduction), 'the the' (Section 3), and 'exchange correlation' (Section 2) should be corrected.
[Section 3] The sentence 'Note because the two Fe atoms on each atomic plane in the G-AFM phase have opposite ms.' is incomplete; the following 'we only list its magnitude' is orphaned. This should be rephrased.
[Section 2] The definition 'The MCA per interfacial area is determined by [E[100]-E[001]]' should clarify that the energy difference is divided by the interfacial area, and define the sign convention clearly.
[Figure 2 caption and Conclusion] The Fig. 2 caption contains a duplicated sentence about k points with large negative or positive contributions. In the Conclusion, the notation 'Lyz,zx' in the matrix element is unclear and should be defined consistently.
[Section 2] The calculation considers only one interface termination (Fe-terminated, O atop Fe) and one film thickness (5 ML). The authors should comment on the expected robustness of the strain-induced switching with respect to termination and thickness, or state that the prediction is specific to this model.
[Section 3] The phrase 'the spin reorientation occurs around 0' should specify units; it should read 'around s=0' (i.e., the lattice-matched condition in the model).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the strain–MCA relation is a parameter-free DFT total-energy difference.

full rationale

The central result — that MCA in the FeRh/MgO bilayer changes from negative to positive as strain goes from -0.5% to +0.5% — is obtained directly from total-energy differences with the spin-orbit coupling included, E[100]-E[001], with no fitted parameters. The strain is an external geometric constraint applied to the slab, and the MCA values are computed outputs. The subsequent k-resolved and PDOS analysis of Fe d-orbital characters is a post-hoc interpretation of the same DFT eigenstates; it is not used as an input to constrain or reproduce the MCA. The DFT inputs (PBE, PAW, VASP) are standard and do not encode the target result. No equation in the paper is shown to reduce to another by construction, and no load-bearing step relies on a self-citation or an imported uniqueness theorem. Concerns about the co-straining of the MgO substrate and the missing Table I/Eq. (1) are physical-realizability and completeness issues, not circularity. The derivation is therefore self-contained with respect to the circularity criteria considered here.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the standard DFT machinery and on the slab model chosen for FeRh/MgO. No parameters were fitted to reproduce the MCA, and the strain values are imposed boundary conditions rather than free parameters. The main unverified premises are the interface geometry, the adequacy of GGA-PBE for MCA in FeRh, and the restriction of the easy axis to the [001]/[100] directions. These are common assumptions in DFT MCA studies but are load-bearing for the specific prediction.

assumptions (4)

domain assumption GGA-PBE exchange-correlation functional accurately describes FeRh structural and magnetic properties
Invoked in Computational details; the text notes LDA fails to give the correct ground state, and GGA is used without a Hubbard U or onsite correction, which can affect MCA magnitudes.
domain assumption The Fe-terminated slab with O atoms atop Fe represents the real FeRh/MgO interface
Used to define the supercell in Computational details; interface termination is not experimentally verified in the paper.
domain assumption MCA equals E[100]-E[001], restricting the easy axis to in-plane or out-of-plane
Stated in Computational details; the calculation does not sample other magnetization directions, so a tilted easy axis would not be captured.
domain assumption Second-order perturbation theory with only dx2-y2 and dxz,yz orbital couplings captures the MCA mechanism
Used in Section 3 and Conclusion to explain the strain effect; other spin channels and orbital pairs are neglected.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Strain controllable magnetocrystalline anisotropy in FeRh/MgO bilayers." pith.science (2026). https://pith.science/paper/QXEP7XB4

@misc{pith2026190804761,
  author       = {Pith},
  title        = {Pith review of: Strain controllable magnetocrystalline anisotropy in FeRh/MgO bilayers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QXEP7XB4}},
  note         = {Machine review of arXiv:1908.04761}
}

read the original abstract

Ultra-thin film of FeRh on insulator MgO substrate has been investigated usingab-initio electronic structure calculations. From this calculation, we have found the interesting effect of epitaxial strain on the magnetocrystalline anisotropy (MCA). Analysis of the energy and k-resolved distribution of the orbital character of the band structure reveals that MCA largely arises from the spin-orbit coupling (SOC) between dx2-y2 andd xz,yz orbitals of Fe atoms at the FeRh/MgO interface. We demonstrate that the strain has significant effects on the MCA: It not only affects the value of the MCA but also induces a switching of the magnetic easy axis from perpendicular to in-plane direction. The mechanism is the strain-induced shifts of the SOC d-states. Our work demonstrates that strain engineering can open a viable pathway towards tailoring magnetic properties for antiferromagetic spintronic applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages

[1]

bcc-like-AFM

IntroductionNowadays, the field of antiferromagnetic (AFM) spintronics has been treated as an promising material in the science community [1-10]. First priority reason is that they do not produce stray fields when it is used to device such as random access memory (RAM) [1-7]. Because structures being similar to their ferromagnetic (FM) counterparts, AFM s...

work page
[2]

Computational detailsWe use the Vienna ab initio simulation package (VASP) to perform the electronic structure calculations. The projector augmented wave formalism is adopted for describing the electron-ion interactions, and plane waves with a kinetic energy cutoff of 350 eV are used to expand the wave functions. The generalized gradient approximation (GG...

work page
[3]

We also list the MCA values and the total energy difference (∆E) between the the G-AFM and FM phases under different strain

Results and discussions In Table I, we list the calculated values of the strain dependence of the spin magnetic moments (∆ms), orbital magnetic moment differences (mo) of the interfacial Fe atom (Fei) and surface Fe atom66(Fes), respectively, for both the G-AFM and FM phases, respectively. We also list the MCA values and the total energy difference (∆E) b...

work page
[4]

Under a compressive strain of -0.5%, the system possesses a large MCA value with the magnetic easy axis being in-plane

Conclusion We have studied the effect of epitaxial strain on the MCA of an ultrathin FeRh/MgO bilayer system by performing ab initio DFT electronic structure calculations. Under a compressive strain of -0.5%, the system possesses a large MCA value with the magnetic easy axis being in-plane. Detailed analyses based on the perturbation theory show this is d...

work page 2016

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

bcc-like-AFM

IntroductionNowadays, the field of antiferromagnetic (AFM) spintronics has been treated as an promising material in the science community [1-10]. First priority reason is that they do not produce stray fields when it is used to device such as random access memory (RAM) [1-7]. Because structures being similar to their ferromagnetic (FM) counterparts, AFM s...

work page

[2] [2]

Computational detailsWe use the Vienna ab initio simulation package (VASP) to perform the electronic structure calculations. The projector augmented wave formalism is adopted for describing the electron-ion interactions, and plane waves with a kinetic energy cutoff of 350 eV are used to expand the wave functions. The generalized gradient approximation (GG...

work page

[3] [3]

We also list the MCA values and the total energy difference (∆E) between the the G-AFM and FM phases under different strain

Results and discussions In Table I, we list the calculated values of the strain dependence of the spin magnetic moments (∆ms), orbital magnetic moment differences (mo) of the interfacial Fe atom (Fei) and surface Fe atom66(Fes), respectively, for both the G-AFM and FM phases, respectively. We also list the MCA values and the total energy difference (∆E) b...

work page

[4] [4]

Under a compressive strain of -0.5%, the system possesses a large MCA value with the magnetic easy axis being in-plane

Conclusion We have studied the effect of epitaxial strain on the MCA of an ultrathin FeRh/MgO bilayer system by performing ab initio DFT electronic structure calculations. Under a compressive strain of -0.5%, the system possesses a large MCA value with the magnetic easy axis being in-plane. Detailed analyses based on the perturbation theory show this is d...

work page 2016